Acceptability of Anal Cancer Screening Tests Among Gay and Bisexual Men. Joshua Thompson

Transcription

1 Acceptability of Anal Cancer Screening Tests Among Gay and Bisexual Men By Joshua Thompson A Master s Paper submitted to the faculty of the University of North Carolina at Chapel Hill in partial fulfillment of the requirements for the degree of Master of Public Health in the Public Health Leadership Program Chapel Hill 2014 Advisor Date Second Reader Date

2 Abstract Objectives. I investigated gay and bisexual men s willingness to self-administer an anal cancer screening test at home. Methods. I reviewed the current literature on acceptability of cytological anal cancer screening tests among gay and bisexual men by searching three databases (MEDLINE, Web of Science, and PsychINFO) and present a qualitative synthesis of the literature. I then performed a secondary data analysis of two national, online surveys of gay and bisexual men: Study I in 2009 with men ages (n=306) and Study II in 2013 with men ages (n=428). I used logistic regression to identify bivariate correlates of willingness to self-administer an anal cancer screening test for both studies and scompared the results. Results. Most men were willing to self-administer an anal cancer screening test (78% Study I; 67% Study II), similar to the findings of the most recent studies of physician-collected tests. In Study I, willingness was higher among men who trusted anal Paps to find treatable cancer (adjusted odds ratio [aor]=1.47; 95% CI 1.04, 2.09) and who believed that men who have sex with men should be screened for anal cancer between one and three years vs. other intervals (aor=2.19; 95% CI 1.17, 4.10). In Study II, willingness was higher among men who perceived greater likelihood of anal cancer (aor=1.57; 95% CI 1.12, 2.20). Their most common concerns were not performing the test correctly and inaccuracy of results. Conclusions. Many gay and bisexual men were willing to self-administer anal cancer screening tests at home. If routine screening is warranted, self-collected home testing could improve participation.

3 Acceptability of Cytological Anal Cancer Screening Among Gay and Bisexual Men A Review of the Literature Introduction For the average man in the United States anal cancer is a rare threat affecting fewer than two men per 100,000 per year. 1 This despite the fact that there has been an almost constant increase in incidence over the past three decades (annual percent change 3-4%). 2,3 Certain subpopulations, such as gay and bisexual men, suffer from a much greater burden of disease, which may account for some of the increase in incidence. 4 The incidence of anal cancer among HIV-negative men who have sex with men (MSM) is approximately 5 cases/100,000 men-years. For MSM living with HIV/AIDs, the incidence of anal cancer is estimated to be between 50 and 130 cases /100,000 men-years depending on whether one uses data from cancer registries or cohort studies. 5-7 Similar to cervical cancer, the majority of anal cancer cases are associated with infection by human papillomavirus (HPV), specifically high-risk types 16 or 18. 8,9 Infection with these high-risk subtypes leads to dysplastic changes that can ultimately result in invasive cancer. 10,11 The prevalence of HPV infection and dysplastic lesions does not decrease with age in gay and bisexual men, in contrast to what is observed in women. 12,13 In fact, a recent study confirmed that the risk of anal cancer among gay and bisexual men increases with age. 7 Prevalence of high-risk HPV types is even higher for MSM living with HIV/AIDS compared to HIV-negative MSM, despite use of antiretroviral therapy. 6,14 Other risk factors (aside from HIV) for persistent HPV infection, subsequent dysplasia, and cancer include the following: having multiple sexual partners, particularly receptive anal sex partners; 6,12-15 smoking; 8,15-17 and low CD4 count. 17,18

4 Anal cytological screening and treatment of dysplastic lesions have been proposed as a way of reducing the burden of anal cancer. 19,20 A recent meta-analysis showed anal cytology has sensitivity and specificity similar to cervical Pap testing (90% and 33%, respectively). 21 No studies have been done to show that routine screening improves mortality rates or other patientcentered outcomes. 22,23 A clinical trial to determine the efficacy of screening is just now beginning recruitment, and it will be several years before the results are published (NCT ). Despite the lack of health outcomes data, some health professionals and organizations have begun recommending screening. 23 The Infectious Diseases Society of America (IDSA) recently updated their primary care guidelines to recommend anal Pap testing for HIV-infected MSM due to the high risk of anal cancer in that population. 24 The new recommendation will likely increase the rate of anal cancer screening in HIV-positive MSM. For this reason, it is important to understand the acceptability of screening among these men. Such knowledge may be invaluable for public health efforts to maximize prevention efforts reach if future trials demonstrate the efficacy of screening. To my knowledge, only one previous review has examined the literature in this area, but the study used a narrative rather than systematic approach, had a broader focus on the general psychological aspects of anal cancer screening, and did not include some more recent articles. 25 The goal of this review, therefore, is to systematically assess what is currently known about gay and bisexual men s acceptance of anal cancer screening. Methods Literature Search Strategy

5 I searched MEDLINE, Web of Science, and PsychINFO for English-language, humanonly studies published from database inception through March 6, I identified additional studies by manually searching the references lists of included studies. Searches combined multiple terms for sexual minority men, anal cancer and dysplasia, and screening or prevention. The search was limited to studies reporting outcomes of the acceptance of anal cancer cytological screening (i.e., anal Pap) or willingness or intent to undergo screening among gay and/or bisexual men or MSM. Inclusion and exclusion criteria are summarized in Table 1 Table 1. Inclusion/exclusion criteria for studies assessing acceptance of anal cancer screening among men. Criteria Category Inclusion Exclusion Population Adult men who identify as anything All other populations other than heterosexual or who report having sex with men Interventions Anal Pap exam (anal cytology) Digital Rectal Exam (DRE) HPV serology Anal HPV testing High-resolution anoscopy Outcomes Willingness or intent to undergo anal cancer screening Beliefs related to anal cancer screening Acceptance of anal cancer screening Time period Studies published through March 6, 2014 Settings Any setting (clinic, community venue, home-testing) Study Design Cross-sectional studies Cohort studies Case-control studies RCTs (HRA) +/- biopsy Epidemiology, natural history, or management of HPV or dysplasia Anal Pap test characteristics Reviews Guidelines Meta-analyses Cost-effectiveness analyses Case reports Public language English All other languages I reviewed abstracts to determine if studies met the inclusion and exclusion criteria. If it was unclear whether a study met criteria, the abstract was marked as included. I reviewed the full text article of all included abstracts against the criteria outlined in Table 1. For studies meeting inclusion criteria, I abstracted relevant data on characteristics of included trials and relevant outcomes using a pre-defined form. Specifically, I was interested in the study design,

6 the sample population, how the study measured acceptance of screening or intent to screen, the overall level of acceptance or intent, and statistically significant multivariate correlates of acceptance or intent. Quality Assessment of Included Studies I evaluated the quality of studies using predefined criteria, based on the Methods for the development of NICE public health guidance, 3 rd edition. 26 I assessed whether the studies adequately described their source and sample population, selection of the sample population, measurement of the primary outcome (e.g. acceptability of screening) and covariates, statistical analyses, overall internal validity of the findings, and the generalizability of those findings to other gay and bisexual men. I assessed risk of measurement bias based only on how the study measured screening acceptability and its covariates, not on other outcomes reported in the trial. I rated each study as being of good, fair, or poor quality. Poor quality studies are those that I judge to have one or more critical flaws, such as a failure to adequately or properly measure screening acceptability (e.g,, measurement bias) or failure to measure important covariates that may affect screening acceptability. Finally, I summarize and qualitatively synthesize the results of the included studies, focusing on overall acceptance of screening or intent to screen, and the multivariate correlates of acceptance or intent to undergo screening.

7 Results Figure 1. Flowchart of study selection process The literature search resulted in 553 unique articles. Figure 1 summarizes results of the search and the study selection process. Ten articles met all criteria and were included in the review. Table 2 summarizes the study design and sample characteristics of each study, as well as how authors measured acceptance of screening. In general, across included trials, there was heterogeneity in study design and methods used to measure acceptance of anal Pap screening. These issues are discussed in more detail in the following sections.

8 Table 2. Overview of study design and quality Screening behavior D Souza et al., (United States) Cohort study Design Study Population Sample Description Multicenter AIDS Cohort Study (MACS) participants Baltimore, Chicago, Pittsburgh, Los Angeles Enrolled all men with a MACS visit during study period n = 1742 (78% of cohort) - unknown % MSM - 47% HIV+ - med. age 55y - 67% White, 20% Black - med. income $40-49,000-23% previous screening Acceptability Measure Acceptance of free, physician-collected Pap Quality Good Gilbert et al., (Canada) Cohort study Men at MSM-frequented venues in Vancouver, British Columbia Randomized time-space sampling n = % gay or bisexual - approx 16% HIV+ - 56% age 25-44y - 73% European or North American ethnic origin - 79% income >$20,000 - approx 24% previous screening Acceptance of free, selfcollected Pap Fair Goodall & Clutterbuck, (UK) Retrospective casecontrol study Single HIV Clinic in Edinburgh Consecutive sampling included all HIV+ MSM attending clinic n = 285 (215 offered screening) - 100% MSM - 100% HIV+ - mean age 39y - unknown ethnic groups, 97% British or White non-british - unknown income, previous screening Acceptance of physician-collected Pap Poor Truesdale & Goldstone, (United States) Retrospective cohort Willingness or intent to get Pap D Souza et al., Cross-sectional (United States) survey Men referred for screening at single surgical clinic with abnormal anal cytology Unknown recruitment procedure, possibly convenience sample? MACS participants Baltimore, Chicago, Pittsburgh, Los Angeles Surveyed entire cohort 2005 n = % MSM - 49% HIV+ - mean age 44y - 78% White, 9% Black - unknown income - 100% screened at least once n = approx 93% MSM - 47% HIV+ - med. age 48y - 63% White, 26% Black - med. income $35,000-11% previous screening Compliance with follow-up recommendations for physician-collected Pap testing Intent to get anal screening in next 6 months Poor Fair

9 Reed et al., (United States) Cross-sectional survey Willingness to repeat Pap after collection Davis, Goldstone, & Cohort study Chen, (United States) National probability sample of self-identified gay and bisexual men 2009 Men and women referred for screening at single clinic specializing in HPV-related, anorectal disease Unknown recruitment procedure 2010 n = % self-identified as gay or bisexual -17% HIV+ - mean age 46y - 81% non-hispanic White, 5% non-hispanic Black - 60% income $60,000-14% previous screening -292 people -97% male, mostly MSM -45% HIV+ - mean age 43 - unknown ethnic groups -63% previous screening Willingness to get physician-collected anal Pap Willingness to repeat Pap after physician-collected test Good Poor Vajdic et al., (Australia) Other Newman et al., (United States) Cohort study Qualitative study Four semi-structured, open-ended focus groups Men presenting or referred to gay health clinic or community-based HIV clinic Consecutive sampling Unknown time period Health care advocates for gay men and MSM in Los Angeles Purposive sampling Unknown year n = % MSM - 64% HIV+ - mean age 45y - unknown ethnic groups, income - 43% previous medical treatment of anus n = men, 3 women - several were gay men of color Willingness to repeat Pap after physiciancollected test Barriers and facilitators of anal cancer screening Poor Fair Botes et al., (Australia) Cohort study MSM = men who have sex with men Unknown source population Consecutive recruitment MACS = Multicenter AIDS Cohort Study n = % MSM - 100% HIV+ - med. age 50 - unknown ethnic groups, income, previous screening Acceptability of Pap after self-collected test Poor

10 Characteristics of Included Studies Sample Populations. Among the 9 quantitative studies, sample sizes ranged from 151 to 1917 men; one study (n = 19) assessed acceptability of screening via focus groups. 35 The most common source populations were individual clinics; four studies reported data from two surgical clinics, two HIV clinics, and one primary care clinic specializing in gay men s health (Table 2). Two studies surveyed men living in large metropolitan areas who were participants in the Multicenter AIDS Cohort Study (MACS). 27,31 Only one study used a national probability sample. 32 Almost all studies recruited samples that were mostly or entirely MSM; though, only one defined criteria for inclusion as MSM, 28 and one study used self-identified sexual orientation rather than MSM status. 32 The proportion of participants who were HIV-positive varied from 16% to 100% of the sample, but most studies (5/9 quantitative studies) had approximately even numbers of HIV-positive and negative men. All quantitative studies included at least some HIV-positive men. The mean or median age of all study samples ranged from Most studies provided little additional detail on the study populations or recruitment procedure aside from the characteristics summarized above. Study Quality. Five of the included studies were of good or fair quality, and the remaining were rated poor quality (Appendix 1). Among the trials rated as poor quality, the most common methodological bias was inadequate description of the study population, and therefore insufficient data to judge the applicability of the results (6/10 studies). Other common methodological flaws included the following: 1) insufficient description of the measurement instruments (4/9 quantitative studies); 2) failure to measure potentially confounding factors

11 associated with acceptability (3/9); and 3) inadequate statistical analysis, specifically not using multivariate modeling to control for confounding among the covariates (5/9). Measurement of Anal Pap Screening Acceptability. There was considerable heterogeneity in study design and how authors measured and defined acceptance of anal Pap screening. Four studies, including two prospective cohorts, one retrospective cohort, and one retrospective casecontrol study, examined actual screening behaviors of men offered or referred for Pap screening Two cross-sectional national surveys measured willingness or intent to get a Pap in the future. 31,32 Two cohort studies assessed participants willingness to repeat screening after undergoing Pap testing. 33,34 I also included one cohort study that assessed acceptability of the Pap after a self-collected test and a qualitative study identifying barriers and facilitators to anal cancer screening. 35,36 For the remainder of the review, I will refer to the outcome measures as acceptability except where it is necessary to differentiate between the studies. Acceptability of Anal Pap Screening. Reported rates anal Pap acceptability varied depending on several factors: 1) the context in which the test was delivered 2) the cost of testing, and if the test was physician- or self-collected. When the Pap was offered for free, acceptability was high (83%-85%). 27,32 Acceptability was lower when the Pap was not explicitly free (12-62%) 29,31,32 or when it was self-collected in a public facility (35%). 28 When measured after men received an anal Pap, acceptability of the test was high and most men were willing to repeat the test in the future. 33,34,36 Factors Associated with Acceptability. Six of the nine quantitative studies measured psychosocial and/or demographic variables associated with acceptance of screening or willingness to screen. Only four created multivariate models of acceptance (Table 3). Multiple studies found that people who worried more about anal cancer or had a history of a previous Pap

12 were more likely to be accepting of the test. D Souza et al. (2008) found that having more sexual partners and being HIV-positive were associated with greater acceptability, 31 but other studies showed no statistically significant association. Similarly, Reed et al. found that people with higher levels of education were less accepting of anal Paps, 32 but D Souza et al. (2013) found no significant association. 27 Studies found opposite associations for both race and income on measures of acceptability. In D Souza et al. (2008), Black participants in the Multicenter AIDS Cohort Study had a greater odds of intent to be screened in the next 6 months compared to White men. In another study of MACS participants 5 years later, Black participants had more than twice the odds of declining an offered Pap compared to White men. 27 Men with low income (<$20,000/year) in Gilbert et al. were more likely to accept a free self-collected Pap, 28 but in Reed et al., men with a higher income ( $60,000/year) were more likely to say they were willing to get a Pap test that cost $ The one qualitative study exploring MSM s beliefs associated with anal cancer screening identified six categories of barriers to screening: lack of awareness of screening, the potential for stigma to be associated with screening, psychological and physical discomfort, the idea of the anus as a private area, overriding concern for HIV, and a general reluctance of men to seek health care. 35 Focus group participants also suggested three possible facilitating factors: increasing the number and diversity of screening sites (including home testing), making changes to the health care system (such as physicians routinely recommending screening), and education campaigns to increase awareness. Discussion

13 I found ten studies that explored the acceptability of anal cancer cytological screening among gay and bisexual men. To my knowledge, this is the first systematic review specifically assessing acceptability of screening in this high-risk population, and the factors that predict acceptance. An earlier review focused on the broader psychosocial aspects of anal cancer screening without regard to a specific population and included three of the studies reviewed here. 25 At first glance, it appears that the fair or good quality studies included in this review found very different degrees of acceptability among gay and bisexual men. Three studies reported that only 29-35% of men were willing to or intended to be screened, 28,31,32 whereas 85% of men in the most recent study accepted a Pap test. 27 Of note, the two studies with the most widely divergent results, D Souza et al and 2013, used essentially the same samples: participants in the Multicenter AIDS Cohort Study. 27,31 Differences in the design and characteristics of the sample populations of these studies may be responsible for the heterogeneous results. Awareness and prior utilization of anal Pap testing were found to be as different between studies as acceptability. In the two earlier studies, D Souza et al. (2008) and Reed et al., only 11 and 14% of men, respectively, had undergone a Pap prior to the study. 31,32 Approximately 25% of men in both of the later published studies had previously been screened. 27,28 Furthermore, Reed et al. reported that only 23% of the men in their sample had heard of Pap testing, 32 but 65% of men surveyed by D Souza et al. (2013) reported being familiar with the test. 27 Since it appears likely that having had a Pap test predicts acceptability, it is not surprising that the study with the highest Pap awareness and proportion of men with a history of Paps also reported the highest acceptability.

14 Two of the studies, D Souza et al. (2008) and Reed et al., measured willingness or intent to be screened, 31,32 and the other two studies, Gilbert et al. and D Souza et al. (2013), measured men s behavior when offered a free screening test and found higher acceptability. 27,28 This runs counter to previous studies that have shown measuring intention generally overestimates actual behavior regarding cancer screening and other preventive health choices Some of this difference may be due to participants explicit or implicit consideration of cost. The behavioral studies offered the tests for no cost to participants. D Souza et al. (2008) asked about men s likelihood to be screened in the near future, without explicitly telling them to consider cost. 31 Reed et al. measured acceptability of a Pap test that cost $150 out of pocket (the primary outcome) and one that was free. 32 When the test was free, 83% of men said they were willing to be screened almost the exact number of men who accepted the free test in D Souza et al. (2013). 27 This suggests that cost considerations are an important part of gay and bisexual men s decision-making process regarding anal cancer screening. Only one study in this review, Gilbert et al., assessed the acceptability of self-collected anal Pap tests. 28 Based on other literature on self-collected tests for sexually transmitted diseases among MSM, I would expect to see high rates of acceptance The anomalously low acceptability reported by Gilbert et al., despite their high proportion of men with previous Pap screening and the offer of a free test, is likely due to differences in setting. This study was conducted within a community-based setting; participants were approached in MSM-frequented venues in Vancouver and asked to self-collect an anal Pap in the venue s bathroom. It is possible that a self-collected test performed at home would be more appealing to men, as suggested by the focus groups in Newman et al. 35

15 I found numerous differences in the magnitude, direction, and statistical significance of the effects of multivariate correlates on men s acceptability of anal Pap testing. Two potentially modifiable factors studies agreed were associated with acceptability are worry about anal cancer and a history of previous Pap testing. Worry, considered a component of risk perception or affective response to risk, is well established as a predictor of preventive health behavior. 43,44 Studies have also shown that previous use of cancer screening tests is associated with greater use in the future. 37,45,46 Different studies in this review reported increased acceptability associated with less education, more sexual partners, and being HIV-positive; however, at least one other included study failed to find a statistically significant association for each one. It is possible that they were simply underpowered to detect the association, but more research is needed to confirm these variables as correlates of anal cancer screening acceptability. The conflicting results regarding income found by Gilbert et al. and Reed et al. may be due to the incentives men received to participate in the studies. 28,32 All men received equal compensation in the Reed study. Although the men in the Gilbert study who answered the survey portion all received the same amount of money, those that agreed to use a self-collected anal screening test received an additional CDN$10. The authors hypothesize that this could be why men with incomes <$20,000 were more likely to accept the Pap. D Souza et al and 2013 finding significant associations with acceptability in opposite directions for Black race is more difficult to explain. 27,31 These studies surveyed the MACS cohort at two different time points but had very different results in terms of overall acceptability and covariates. It is possible that the difference is due to some change in the Black participants over time, but a more likely explanation may be that the difference is due to how

16 each study asked men about screening acceptability. The earlier study asked men about the chance they would be screened in the future, and the latter offered men an actual anal Pap. Predictive factors for intent and behavior are not always the same, 37 and it is possible that Black men overestimate the actual chances they will use anal cancer screening more than White men. Limitations of the Literature The current body of literature on gay and bisexual s men acceptability of anal cancer cytological screening has significant limitations. Only three good quality, multicenter or nationally representative studies have explored this topic, and they had mixed results. Two of those studies measured men s intent to be screened with a cross-sectional survey, rather than assessing actual behavior. They may not represent what could be expected if routine screening was recommended for MSM. The majority of studies included in this review had samples limited to single clinics and many failed to adequately describe their clinic populations. This makes it difficult to determine the applicability of their findings to other men. Limitations of This Review The scope of this review was limited to only three databases and only articles published in English. A more thorough search that included additional databases and the grey literature may find other relevant studies. Furthermore, only a single reviewer read the abstracts and full text articles to determine inclusion. This may introduce some error. Future Directions Now that the IDSA is recommending screening for some high-risk individuals, future studies should focus on determining screening uptake, rather than intent to be screened, among men referred for screening and the sociodemographic and psychological factors that predict use of screening. Although all MSM have an increase chance of developing anal cancer, behavioral

17 research should focus on MSM living with HIV due to their uniquely high risk and the uncertainty regarding screening efficacy, especially in the HIV-negative population. Future research should also attempt to confirm the relationship of potential covariates to screening behavior that previous studies have disagreed upon, such as number of partners and HIV status. Lastly, more work should be done to sample MSM populations in a way that allows bisexual men to be assessed as a distinct group. Bisexual men suffer from a number of health concerns and health access disparities independent of gay men, but most of studies in this review (and in LGBT research generally) combined the two populations due to small samples. 47 Conclusion In conclusion, the literature suggests that anal cancer cytological screening is acceptable to gay and bisexual men when offered at no charge or as part of a study but may be less acceptable when real-world costs are considered. Efforts to improve screening uptake (if screening is proven to be efficacious in preventing anal cancer) should target men with no prior history of screening and concentrate on informing men at the highest risk for anal cancer about the disease and prevention options.

18 12% likely to be 35% accepted Pap n.s Table 3. Summary of results for studies that reported multivariate correlates Multivariate Correlates of Acceptance, OR (95% CI) Author Primary Outcome Year (n) Race a Worry about Income Education # Partners Previous Pap HIV+ anal cancer D Souza et al (1.4, 2.5) b 2.9 (2.2, 3.9) c 1.9 (1.4, 2.6) 1.9 (1.5, 2.5) 2008 (1917) 31 screened in next 6 mo (1.06, 1.90) 17% possibly will be 4.7 (3.3, 6.7) d screened Gilbert et al (766) 28 (1.24, 3.22) e (1.05, 2.67) Reed et al. 31% willing to get Pap n.s. n.s (306) 32 (1.18, 3.98) f (0.32, 0.89) g (1.06, 2.72) h D Souza et al (1742) 27 15% declined Pap 2.15 (1.18, 3.9) n.s. n.s n.s. n.s. = not statistically significant a Black vs White b 3 receptive anal male sex partners in past 6 months vs none c Moderately concerned vs not concerned d Very concerned vs not concerned e <$20,000 vs >$20,000 per year f $60,000 vs <$60,000 per year g Greater than vs less than high school education h 4-point scale, from not at all (coded as 1) to quite a lot (coded as 4)

19 Appendix 1. Quality Assessment of Studies of Anal Cancer Screening Acceptability Among Men Who Have Sex With Men First author, year Eligibility criteria clearly described? Subjects representative of the overall source population? Generalizability to other gay/bisexual/ MSM pop.? (high/moderate/ low) Measured covariates based on a theoretical model or prior research? Outcome measures valid and reliable? Created multivariate model of covariate to adjust for confounding? Quality rating (good/fair/ poor) Comments D Souza et al, Yes Somewhat Moderate Yes Yes Yes Good Men offered Pap more likely to be White, have a college degree, have income >$40,000, have 0 receptive sex partners in past 6 mo. than those not offered Pap. MACS cohort only from 4 large urban centers. Pap was offered free likely would not be in real world setting. Outcome was acceptance of Pap offer Reed et al., Yes Yes Moderate to High Yes Yes Yes Good National probability sample, but participants were mostly high-income and White and selfidentified as gay or bisexual Collapsed 5-point Likert response for willingness into dichotomous outcome Performed 2 rounds of pretesting prior to study start D Souza et al, Yes Yes Moderate Yes Unclear Yes Fair Included all men enrolled in Multicenter AIDS Cohort Study, which has been described in detail in previous articles. Men only came from 4 large urban centers Question used to measure intent to seek screening for anal cancer also asked about screening for anal warts and other anal health problem[s]. This may overestimate actual cancer screening intent.

20 First author, year Eligibility criteria clearly described? Subjects representative of the overall source population? Generalizability to other gay/bisexual/ MSM pop.? (high/moderate/ low) Measured covariates based on a theoretical model or prior research? Outcome measures valid and reliable? Created multivariate model of covariate to adjust for confounding? Quality rating (good/fair/ poor) Comments Gilbert et Yes Yes Moderate No Yes Yes Fair Source population were men attending MSMfrequented al., venues likely not representative of men who do not attend such venues. Did not include several important covariates in multivariate model: orientation, smoking status, HIV status potentially due to concern of colinearity with venue-type Outcome was acceptance of Pap offer Newman et No Unclear Low N/A N/A N/A Fair Qualitative study. al., Focus group participants were primarily health care advocates rather than gay or bisexual or MSM patients Botes et al., No Unclear Unclear No No No Poor Did not describe source or sample population. Survey response options not mutually exclusive. Did not define acceptance. No bivariate or multivariate analysis of covariates. Covariates did not include any demographic or psychological variables Reported correlation coefficients rather than ORs. Davis et al., No Unclear Unclear N/A Unclear No Poor Did not describe recruitment procedure. Did not describe source or sample population. Did not measure any covariates. Exposure to multiple screening modalities at same visit (digital rectal exam, anoscopy) may confound outcome. Outcome was whether discomfort from Pap would prevent from repeating.

21 First author, year Eligibility criteria clearly described? Subjects representative of the overall source population? Generalizability to other gay/bisexual/ MSM pop.? (high/moderate/ low) Measured covariates based on a theoretical model or prior research? Outcome measures valid and reliable? Created multivariate model of covariate to adjust for confounding? Quality rating (good/fair/ poor) Comments Goodall & Clutterbuck, Yes Unclear Low No Yes No Poor Sample included all HIV+ MSM at single clinic over one year Did not say why some patients (75%) were offered screening and others were not Outcome was acceptance of Pap offer. Reported percentages and p-values, not ORs no multivariate analysis Truesdale & Goldstone, Yes Unclear Low Yes Yes No Poor Unknown recruitment procedure. 12% response rate for contacted lost to followup patients Results represent a single clinic, mostly White and well-educated. No multivariate analysis. Vajdic et al., Yes Unclear Low N/A Unclear No Poor Consecutive sampling of men from gay health clinic and HIV clinic unlikely to be generalizable Did not measure any covariates of preparedness to repeat Pap test. Reported acceptance rating (5-point Likert scale) as mean and range limited ability to interpret results No comparisons made of acceptance based on different covariates other than blind vs. anoscope-guided sampling

22 Gay and Bisexual Men s Willingness to Use a Self-Collected Anal Cancer Screening Test Introduction Incidence of common cancers affecting men in the United States (prostate, lung, and colorectal cancer) have declined over the past decade. 48 Anal cancer incidence, however, has increased over the past three decades, and currently 1.5 men per 100,000 are affected per year. 48,49 This increase is attributable, at least in part, to high incidence rates among men who have sex with men (MSM; ~5 men per 100,000 per year). 49 Rates are even higher among MSM living with HIV/AIDS, with recent estimates ranging from 65 to 131 men per 100,000 per year. 7,50 The majority (>70%) of cases of anal cancer are associated with human papillomavirus (HPV) infection, specifically with high-risk type 16 or Risk factors for new or persistent anal HPV infections, as well as for anal cancer, include receptive anal intercourse, having multiple sexual partners, and smoking. 8,12,52-55 These risk factors may explain why high-risk HPV types are common in HIV-negative and HIV-positive MSM (12.5% and 35.4%, respectively, for type 16). 50,56,57 Anal cytological screening and treatment of dysplastic lesions could represent an important strategy for preventing anal cancer. 19,20 Routine cervical Papanicolaou (Pap) testing has resulted in a dramatic decrease in the incidence of invasive cervical cancer in women over the last 50 years. 58 It is possible that anal cancer screening could have similar effects for MSM, given the similarities between cervical and anal cancers and studies showing anal Pap tests have similar accuracy to cervical Paps. 21 No studies have been conducted, though, to determine anal cancer screening s effect on mortality or other patient-centered outcomes. 59 Despite this

23 uncertainty, some clinicians are now recommending high-risk men be screened for anal cancer because of the high incidence of anal cancer. 60 The Infectious Disease Society of America recently recommended screening all HIV-positive MSM for anal cancer with anal Pap tests. 24 Studies of anal cancer screening behavior among MSM have focused almost exclusively on men s acceptance of physician-collected Pap tests. 25,27,32 Self-collected anal Paps have been shown to have accuracy similar to clinician-collected tests. 61,62 Studies of self-tests for HIV and other sexually transmitted diseases have found high rates of acceptability among MSM, and it is possible that allowing men to use a self-test for anal cancer would increase screening uptake The purpose of the present study was to examine gay and bisexual men s willingness to use a self-collected anal cancer screening test at home and to identify correlates of willingness to use a self-test. Findings could help inform future programs for promoting anal cancer screening among this higher-risk population Methods I used data from two separate research studies of gay and bisexual men. The studies are described in detail elsewhere and briefly here Study I Sample and procedures. The sample for Study I came from an existing, national panel of US households maintained by Knowledge Networks (Palo Alto, CA) Panel members were recruited using list-assisted random-digit dialing. Panel members received free Internet access or small cash payments in exchange for completing multiple online surveys each month. Study I was limited to men aged and oversampled for self-identified gay and bisexual men. Of the

24 874 eligible panel members invited to participate, 609 (70%) completed the survey in January Of those who completed the survey, 306 men self-identified as gay (n = 236) or bisexual (n = 70) and were included in the current analysis. The Institutional Review Board at the University of North Carolina approved this study. Measures. The Study I survey is available online at Brewer et al. developed the survey items based on their previous work on HPV-related diseases They cognitively tested the survey to refine item design with 28 gay and bisexual men and performed a second round of testing with 8 additional men prior to beginning the study. My primary outcome was willingness to self-administer an anal cancer screening test at home. Introductory text to the section on anal cancer screening read, Doctors can use an anal Pap test to identify anal cancer. An anal Pap test is when a doctor collects cells from the anus using a swab (like an extra long Q-tip) and examines them for changes. An anal Pap test is not the same as a test for anal gonorrhea, a colonoscopy, or a digital rectal exam. The survey assessed willingness to get a physician-collected anal Pap test under two conditions: (1) if it were free and (2) if it cost $150 out of pocket. Then the survey assessed willingness to self-administer an anal cancer screening test: How willing would you be to use an anal swab on yourself to screen for anal cancer? You would do this test at home by yourself instead of going to a doctor s office. Response options for willingness items were definitely wouldn t, probably wouldn t, not sure, probably would, and definitely would. I collapsed responses into two categories: willing (probably or definitely would) and not willing (all other responses). The survey assessed participants awareness of HPV prior to the study and knowledge of HPV using five factual items. I coded three to five correct responses as high HPV knowledge and the rest as low HPV knowledge, based on the median number of correct responses for the

25 entire sample. The survey also measured respondents perceived knowledge of several HPVrelated diseases (genital warts, oral cancer, and anal cancer), as well their worry about the disease, perceived likelihood of developing it, and belief that having the disease would change their lives. Because pilot testing showed men had low familiarity with HPV-related disease, I included brief informative statements about each disease prior to asking questions about them. Brewer et al. measured several demographic and health-related characteristics as potential covariates. For urbanicity, they defined urban as living in a metropolitan statistical area based on zip code. 69 The survey also asked about history of sexually transmitted infections, genital warts, lifetime number of sexual partners, number of partners in the past year, and whether participants had disclosed sexual behavior with men to their physician. Study II Sample and procedures. The Study II sample was drawn from the Harris Interactive LGBT Panel, a subset of the Harris Poll Online Panel (Rochester, NY). The Harris Poll Online Panel is a voluntary research panel that includes members throughout the entire country; US participants are similar to the general population for several demographic characteristics. 70 Panel members complete multiple surveys each month and receive points that can be exchanged for rewards. Eligibility criteria for Study II included being age (because the primary focus of the study was on HPV vaccination among young adults), living in the US, and self-identifying as lesbian, gay, bisexual or transgender. Of 2,014 panel members eligible for the study, 1,005 (50%) consented to and completed an online survey in October or November Of those who completed the survey, the 428 men who self-identified as gay (n = 309) or bisexual (n =119)

26 were included in the current analysis. The Institutional Review Board at The Ohio State University approved the current study. Measures. Reiter et al. based the Study II survey on the survey used for Study I, as well as subsequent surveys on HPV vaccination My primary outcome was willingness to selfadminister an anal cancer screening test at home. The survey item read, There is also a home test that may help screen for anal cancer. This test would involve using a swab (like a Q-tip) to get an anal specimen. You would do this test at home by yourself instead of going to a doctor s office. How willing would you be to use an anal swab on yourself to screen for anal cancer? As with Study I, the five response options were collapsed into willing to self-test (probably or definitely willing) and not willing (note sure, probably or definitely not willing). An additional item asked participants to select potential concerns they had about using a home test for anal cancer from a list of predefined response options. The survey assessed participants awareness of HPV prior to the study and knowledge of HPV using five factual items. I coded four or five correct responses as high HPV knowledge and less than four correct responses as low HPV knowledge, based on the median number correct for the entire sample. The survey also measured respondents worry about and perceived seriousness of HPV-related disease, as well as their perceived likelihood of developing specific diseases (genital warts, oral cancer, and anal cancer). Reiter et al. measured several demographic and health-related characteristics as potential covariates. For urbanicity, they defined urban as self-report of living in an urban or suburban area. The survey also asked about history of sexually transmitted infections, genital warts, lifetime number of sexual partners, number of partners in the past year, and whether participants had disclosed their sexual orientation to their physician.

27 Analyses I analyzed data from the two studies separately given their different samples. I used logistic regression to identify bivariate correlates of willingness to self-administer an anal cancer screening test. I entered statistically significant (p <.05) bivariate correlates into multivariate logistic regression models for each study. I used McNemar s test to compare Study I participants willingness to self-administer an anal cancer screening test to their willingness to get a physician collected anal cancer screening test that cost $150 or was free. I used Pearson s chi-squared test to compare the two study samples willingness to self-administer the test. I analyzed data with Stata release 13 (StataCorp LP, College Station, TX). All statistical tests were 2-tailed, using a critical α = Results Participant Characteristics Most men in Study I identified as gay (77%) and were HIV-negative (83%) (Table 1). Their median age was 47 years (range 20-59), and 96% were older than 26. Most were non- Hispanic White (81%), had a college degree (56%), were insured (86%), and lived in an urban area (93%). Only 23% had ever heard of an anal Pap before the survey, and 14% reported having had an anal Pap in the past. Most men in Study II identified as gay (72%), and almost all were HIV negative (96%). Their mean age was 23 years (range 18-26). Most identified as non-hispanic White (64%), had

28 health insurance (80%), and lived in an urban area (84%). Fewer than half had a college degree (47%). Willingness to Self-Administer an Anal Cancer Screening Test Most men in Study I (78%; 95% confidence interval [CI] 73%, 82%) were willing to selfadminister an anal cancer screening test at home. Men were somewhat less willing to selfadminister the test than to get a free physician-collected test (78% vs. 83%; χ 2 = 4.41; p =.04). They were much more willing to self-administer the test than to get a physician-collected test that cost $150 (78% vs. 31%; χ 2 = , p <.001). Two-thirds of Study II men (67%; 95% CI 62%, 71%) were willing to self-administer an anal cancer screening test at home, a lower percentage than among Study I participants (p <.001). The most common concerns about self-administrating the test that participants cited were I might not do the test right (61%) and the test might not be accurate (59%) (Figure 1). About 29% of the men said they would rather go to a doctor to get screened for anal cancer, 22% were concerned the test would hurt, and 17% thought it would be embarrassing. Correlates of Willingness In Study I, men s willingness to self-administer an anal cancer test was higher if they believed MSM should be screened between one and three years (87%) rather than screening at other intervals (adjusted odds ratio [aor] = 2.19; 95% CI 1.17, 4.10) in multivariate analysis (Table 2). Willingness was also higher among men who agreed more with the statement: If I got regular anal pap tests, I would trust them to find anal cancer when it is still treatable (aor = 1.47; 95% CI 1.04, 2.09). Additional correlates of willingness in bivariate analysis were higher

29 worry about anal cancer, higher knowledge about HPV, belief that HIV affects a man s chances of getting anal cancer, being HIV-positive, and being non-hispanic white when compared to other race/ethnicity (all p <.05). In Study II, willingness to self-administer an anal cancer screening test was higher among men who had higher perceived likelihood of developing anal cancer (aor = 1.57; 95% CI 1.12, 2.20) in multivariate analysis (Table 3). In bivariate analyses, correlates of willingness to use a self-test were higher worry about diseases caused by HPV, higher knowledge of HPV, having five or more lifetime sexual partners, having a college degree, and being older (all p <.05). Discussion In the two national samples, most gay and bisexual were willing to self-administer an anal cancer screening test at home. The proportion of men willing to use a self-collected test was similar to or higher than those previously reported for physician-collected Pap tests. 27,31,63 The possibility that men may prefer a self-collected home test over one performed by a physician is supported by my findings that Study I men were more willing to use the self-collected test than to get a physician-collected Pap that cost $150 and that fewer than a third of Study II participants stated a preference for a physician-collected test. Of note, I found much greater willingness to use a self-collected anal cancer screening test than a previous study of acceptability among MSM. Gilbert et al. reported only 35% of men were willing to use a self-collected test, but their sample was limited to men attending MSM-frequented venues and the test was collected in venue bathrooms. 28 The increased privacy, comfort, and safety of collecting the test at home may be an important component of men s willingness to self-administer anal cancer screening. This

30 relationship has been suggested previously by focus groups and studies of the acceptability of HIV self-testing. 35,75 I identified three potentially modifiable correlates of willingness to use self-collected testing in my multivariate analyses. Men in Study II who perceived they had a higher likelihood of anal cancer were more willing to self-administer an anal cancer screening test. Reed et al. previously identified this association in their analysis of willingness to get a physician-collected Pap test. 63 Multiple health behavior studies have established that risk perceptions, including perceived likelihood, are important predictors of cancer screening participation and other health protective actions. 44,76,77 This study extends that link to the use of self-administered cancer screening tests. Men s willingness to use a self-collected test in Study I was associated with their trust in anal Paps to find cancer when it is still treatable. Belief in the efficacy or benefits of screening is another construct frequently correlated with participation in cancer screening programs. 78,79 However, Reed et al. did not find a statistically significant association between perceived test efficacy and willingness to get a physician-collected Pap. 32 Similarly, in a survey of HIVpositive and HIV-negative MSM enrolled in the Multicenter AIDS Cohort Study, D Souza et al. found that belief in the utility of anal Pap tests was associated with acceptance of a free test in bivariate analyses, but the association was not significant in their multivariate model. 27 It is possible that test effectiveness is more important to men when considering a home-based test, as a test performed by a physician may be assumed to have some minimal level of effectiveness. This idea is supported by qualitative work showing some people worry that home-based STI tests are less accurate and my own findings that the majority of men in Study II had concerns about the self-collected test s accuracy and their ability to perform it correctly. 80